Causal logic

This paper proposes a logic for causal reasoning inspired by the idea of an event tree. Event trees provide a natural and familiar framework for probability and decision theory, but they lack the modularity that would make them convenient models for a logic. Our logic overcomes this deficiency by generalizing event trees to event spaces. The situations (nodes) in an event tree are related to each other primarily by precedence; one happens before another. The situations in an event space correspond to nodes in variously detailed event trees and are therefore related by relations of refinement and implication as well as precedence. A causal logic that uses event spaces as its models can therefore combine ideas and intuitions from numerous sources, including probability theory, temporal logic, the situation calculus, and situation semantics.